Rough-Cut Capacity Check: See the Overloaded Work Center Before You Commit the Plan
Compare your planned production load (hours of work) against the available capacity (hours) of each key work center, week by week, so the overloaded bottleneck lights up red before you commit the schedule — not after you've already promised the order and missed it. A planner reviews the load-vs-capacity picture and approves, or sends it back for leveling, before the plan is published.
A logged-in tool where you import the production plan, the standard hours per item per work center, and the available hours per work center, the tool buckets everything into weeks, computes load percent and over/under hours for each work center each week, highlights the overloads, the planner approves or sends it back for leveling, and you export a clean capacity report CSV.
Before you start
- A Supabase account (free)
- A Vercel account (free)
- A Resend account (free)
- Your production plan (items x quantity, with a date or week)
- Standard hours per item per key work center (CSV or Sheet)
- Available hours per work center per week (CSV or Sheet)
- Claude Code or any AI coding agent
The problem this kills
You build the production plan, it looks fine on the sheet, and you commit. Three weeks later the press line is buried, jobs are stacking up behind one machine, and you're calling customers to push dates you already promised. The overload was sitting in the numbers the whole time — you just couldn't see it until the floor was already on fire.
Rough-cut capacity planning is the simple, powerful check that catches this early: for each key work center, add up the hours of work the plan demands (each item's quantity times its standard hours on that work center) and compare it to the hours that work center actually has that week. If the demand is 220 hours and the work center has 160, you're at 138% — that's your bottleneck, and you know it before you commit instead of after you miss.
Most teams either skip this entirely or do it in a sprawling spreadsheet with pivot tables that one planner babysits. It breaks quietly: a week gets double-counted, standard hours go stale, a new item has no routing and silently contributes zero load, and the "approved" plan is really just the last version someone happened to save. When the check is wrong or skipped, the bottleneck stays invisible until it's a missed shipment.
What you'll build
A simple internal web app for your planning team. You import three things you already have: your production plan (item, quantity, and the week or date it's due), the standard hours per item per key work center (how long one unit takes on each machine or line), and the available hours per work center per week (capacity after shifts, maintenance, and downtime).
The tool aggregates everything into weekly buckets, computes the load on each work center each week (sum of quantity times standard hours), compares it to available hours, and shows you utilization percent and over/under hours — with every overloaded work center clearly highlighted. The planner reviews the load-vs-capacity picture, sees exactly which work centers blow past 100% and in which weeks, and either approves the plan or sends it back for leveling before it's published. Then you export a clean capacity report CSV. Items with no standard hours are surfaced as warnings, never silently treated as zero load.
What's inside the Implementation Plan
The downloadable plan is a step-by-step file you paste into an AI coding agent. It opens by interviewing you about your business — which work centers actually matter (the few real constraints, not every machine), how your plan and routings are named and exported, whether you bucket by week or some other period, how you measure available hours (shifts, efficiency, planned downtime), what counts as an overload worth flagging, and the messy exceptions (items with no routing, shared work centers, multi-week jobs, partial weeks). It reflects a short tailored spec back to you and gets your thumbs-up before it builds anything, so the check matches how your plant actually plans — not a generic template.
From there it walks the agent through the data model, the three imports with duplicate guards, the weekly bucketing and load calculation, the utilization-and-overload view, the planner approval gate, and the capacity-report CSV export. Every step ends with a ready-to-copy prompt. There's a full "No API yet?" path that uses Google Sheet / CSV imports as the data source and produces a clean capacity-report CSV in the columns your downstream tools expect — so you can build and run the whole thing this afternoon, no matter what ERP or scheduling system you're on.
The governance it includes (this is the point)
A committed production plan drives what your floor builds and what you promise customers — so the controls aren't optional. The plan builds in login so only your planning team can use the tool, row-level security so you only ever see your own plant's plan and capacity, a complete audit trail of who imported what and who approved which plan, a hard human-approval gate so no plan is published until a planner reviews the load-vs-capacity picture and signs off (or sends it back for leveling), and duplicate guards so the same work-center-and-week can't be counted twice when you re-import.
Who it's for
Production planners, plant managers, and anyone who sits in the sales & operations planning (S&OP) meeting and has to answer "can we actually do this?" If you can name your handful of bottleneck work centers and tell me how you measure their hours, you can build this — and you'll stop committing plans that were overloaded from the start.
You've got this — start with the plan, paste the first prompt, and answer the interview. You'll have your first load-vs-capacity chart, with the bottleneck lit up red, on screen this afternoon.